High-Flow Oxygen in Reducing Shortness of Breath Caused by Exercise in Patients With Cancer: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for High-Flow Oxygen in Reducing Shortness of Breath Caused by Exercise in Patients With Cancer (NCT02357134), a randomized phase 2 study. The primary outcome is Modified Borg Scale Dyspnea Intensity at Isotime.

2. Clinical Question

3. Trial Design

This is a randomized phase II trial with a parallel design and double masking, according to the registry record. It evaluates four exercise-session conditions: high-flow oxygen, high-flow air, low-flow oxygen and low-flow air. The high-flow conditions use nasal prongs; the low-flow conditions use a nasal cannula. All are described in relation to a structured stationary bicycle exercise session. Which people were masked is not reported in the registry record.

The registry record lists enrollment as 74. The primary outcome description says data are provided for 44 participants who completed the study. The record does not explain the relationship between those counts, give arm-specific completion details, or describe how participants were assigned among the arms. Allocation ratios and the method used to generate the random sequence are not reported in the registry record.

The listed study start date is 2015-03-17, and the primary completion date is 2020-12-30. The study status is recorded as unknown. M.D. Anderson Cancer Center is identified as the sponsor. The study is described as a phase II trial; further details about its design objectives or the rationale for the selected design are not reported in the registry record.

The primary outcome is Modified Borg Scale Dyspnea Intensity at isotime, assessed at the end of the second and third exercise tests. Listed secondary outcomes include dyspnea unpleasantness over time during the third constant-work exercise test, leg discomfort, exercise endurance and adverse events. The registry record also lists quality-of-life assessment and questionnaire administration among the interventions or activities. Their timing and role in the analysis are not reported in the registry record.

4. Participant Flow and Accounting

The registry reports how participants moved through each stage of the study. All counts below are exactly as posted; missing counts are not inferred from other stages.

Before assignment: A total of 74 participants were enrolled but 45 patients were randomized for this study. 29 patients were not randomized for various reasons (became ineligible or too symptomatic to participate while waiting=14, scheduling conflict=9, exertional dyspnea too low=4, could not complete baseline test due to desaturation=1, hypertension during exercise=1)

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: The data is provided for 44 participants who completed the study.

5. Baseline Characteristics

The table reproduces the baseline characteristics posted in the registry, which describe the participants analysed at baseline and help show whether the randomized groups were comparable.

Population: The data is provided for 44 participants who completed the study.

6. Primary Endpoint

The registry lists the primary outcome as Modified Borg Scale Dyspnea Intensity at Isotime, measured over End of second exercise test and end of third exercise test, approximately up to 12 minutes.

The primary endpoint is the mean difference in Modified Borg Scale Dyspnea Intensity at isotime between the second and third exercise tests. The second test used low-flow oxygen for all patients; the third used the assigned intervention. Isotime denotes the common exercise-time point used for the comparison. The record describes assessment at the end of each test and gives an approximate timeframe of up to 12 minutes.

The Modified Borg Scale is a subjective measure of breathlessness intensity, with values ranging from 0, described as no shortness of breath, to 10, described as the worst possible shortness of breath. A continuous score can retain gradations in reported symptom intensity rather than reducing responses to a simple event count. The mean summarizes the group’s average score difference, while the confidence interval conveys uncertainty around that estimate.

Using the second test as the reference provides a within-participant exercise-test comparison, while the third test reflects the assigned condition. However, the registry description does not specify the subtraction order used to form the difference. The sign of an estimate therefore cannot be confidently translated into the direction of change between tests from this description alone. The record does not give further detail on how isotime was selected or how scores were collected.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

What does a mean difference describe?

The primary outcome is continuous: it is a score on the Modified Borg Scale, rather than a count of events or a time-to-event measure. A mean difference summarizes the average contrast in scores as defined by the analysis. In this trial, the registry describes the contrast as being between the second and third exercise tests. It does not state which test’s score was subtracted from which, so the sign’s direction is not fully interpretable from the record.

A mean is useful for describing the center of a group’s score differences, but it does not show how scores vary among individual participants. The posted confidence interval adds information about statistical uncertainty in the estimated mean. It is not a range that contains every participant’s score or every participant’s change. The interval’s interpretation depends on the analysis assumptions, which are not described in detail in the registry record.

How should the confidence intervals be read?

A confidence interval expresses the uncertainty associated with an estimate under a statistical procedure. Wider intervals generally indicate less precision than narrower intervals, all else being equal. The intervals posted for the four arms convey uncertainty about each arm’s mean difference. They do not directly quantify uncertainty about a contrast between two arms. Such a contrast requires its own estimate and uncertainty calculation; those comparisons are not posted for this endpoint.

Whether an interval includes zero can help describe which values remain compatible with the estimate and its uncertainty under the method used. It should not be treated as a complete interpretation of the evidence. The confidence level, the analysis model, and any adjustments or other inferential procedures matter. The record identifies the dispersion as a 95% confidence interval, but further analytical details are not reported in the registry record.

What is the role of randomization?

Random assignment is intended to support fair comparisons by reducing systematic differences between groups at the start of a study. Its value depends on how allocation was conducted and on how participants are handled in the analysis. The record labels the trial randomized, but does not describe sequence generation, allocation concealment, or the analysis population. It is therefore not possible to assess those features from the registry record alone.

The four-arm structure allows the study to examine different combinations of flow and gas during exercise. Yet the posted endpoint estimates are arm-specific, and the record does not provide estimable two-arm comparisons. It would be inappropriate to infer the statistical result of a contrast by comparing whether separate arm confidence intervals overlap. Separate intervals are not a substitute for a direct comparison with its own uncertainty.

What does the planning illustration represent?

The record describes a sample-size planning illustration based on the observed effect, with a two-sided significance level of 0.05, equal allocation between two arms and a normal approximation. The effect to detect is set as a stated fraction of the observed effect. Such calculations are planning illustrations: they show what sample sizes would follow under those assumptions. They are not the observed power of this trial, and they do not show that the original trial was designed using those assumptions.

The registry record does not report the trial’s own sample-size justification or statistical plan. It also does not report the reasoning behind the planning assumptions. The illustration cannot answer whether the study was adequately sized for its intended questions, especially when the posted primary outcome data cover only the stated completer population and direct comparisons are unavailable.

Why can a continuous outcome be difficult to compare?

Continuous scores often vary more than a simple statistical model expects, a feature called over-dispersion. In general, over-dispersion can make uncertainty estimates too small if it is not accounted for, which can make results appear more precise than they are. The registry record does not report whether over-dispersion was assessed or how it would be handled. It also does not describe the distribution of individual scores, model diagnostics, or any transformations.

Repeated exercise tests raise further analytical questions in general. Scores from the same participant are related, and a method may need to account for that dependence when using observations across tests. The registry record does not describe whether the analysis used a repeated-measures model, a paired comparison, or another approach. These details cannot be inferred from the displayed arm-level means and confidence intervals.

10. Multiplicity, Missing Data and Analysis Populations

The registry record lists a primary outcome and four secondary outcomes. Considering multiple outcomes can raise a multiplicity issue: when several questions are tested, the chance of at least one apparently notable result may increase if no adjustment or prespecified testing strategy is used. Whether multiplicity was addressed, and how, is not reported in the registry record. The posted arm-level estimates do not include direct comparison p-values or a stated multiplicity procedure.

The record lists zero analyses. That does not explain whether analyses were conducted outside the registry record or whether results were withheld; the reason is not reported. A statistical analysis plan, including planned contrasts, model specification, covariate adjustment and handling of repeated measurements, is not reported in the registry record.

The enrollment entry is 74, while the primary outcome description says data are provided for 44 participants who completed the study. These counts describe different aspects of the record; no reconciliation, participant flow, or arm-specific account of missing endpoint data is reported. The record does not describe reasons for non-completion, whether missing scores were related to symptoms or treatment assignment, or whether any imputation or other missing-data method was used.

The analysis population is not named beyond the statement that the primary outcome data concern participants who completed the study. Whether the analysis included all randomized participants, only those who completed both tests, or another defined population is not reported in the registry record. The extent to which these choices could affect the estimates cannot be assessed from the posted information.

11. What the Result Does, and Does Not, Mean

The posted results describe mean differences in Modified Borg Scale Dyspnea Intensity at isotime for each of the four exercise conditions, along with 95% confidence intervals. They indicate the precision of the arm-specific estimates for the participants represented in the outcome data. The intervals for the two oxygen arms lie below zero, and the intervals for the two air arms span zero. Because the subtraction order is not specified, the negative sign cannot be confidently translated into a direction of change between tests.

These estimates do not establish that one condition is better than another. Direct two-arm comparisons for the continuous endpoint are missing or not estimable in the registry record. Nor do the arm-specific confidence intervals establish that the interventions caused a change: the analysis details and participant handling are not reported, and the second and third tests differed in assigned condition as well as their position in the testing sequence.

The results do not show how an individual participant’s breathlessness changed, whether the average score difference corresponds to a meaningful change for patients, or whether any difference persisted beyond the exercise tests. A group mean can conceal varied individual responses, and no threshold for a clinically important score difference is reported in the registry record.

The registry also lists adverse-event fields with zero values across all arms, including zero values for the numbers at risk and affected. Since the recorded at-risk values are zero, those fields cannot establish that no participants experienced adverse events. The record does not provide an explanation for these entries or a usable account of adverse-event incidence. These posted results therefore support a limited statistical description, not a complete assessment of benefit, harm or clinical importance.

12. Why This Trial Matters Statistically

This trial’s statistical structure is relevant because it combines a subjective continuous symptom score with repeated exercise testing and four intervention conditions. The endpoint compares scores from a common second-test condition with scores from a third test under assigned conditions. That makes the definition and direction of the difference important: without a clear subtraction order, even the sign of the reported estimate has an ambiguous interpretation.

The randomized design may support comparisons among conditions, but the posted arm-level results do not include estimable two-arm contrasts. This distinction is central to interpreting multi-arm trials. Descriptive estimates within each arm can inform what was observed in that arm, but they do not answer whether arms differ. A reader needs the direct contrast and its uncertainty to assess that question.

The record also illustrates why endpoint populations and missingness matter. Enrollment is listed as 74, while the primary outcome data are described for 44 study completers. Without a participant-flow account or an analysis description, it is difficult to assess how representative the posted estimates are of the randomized participants. The limitation is about interpretability, not an assumption about why the counts differ.

Finally, a planning calculation based on an observed effect should not be confused with the trial’s original design justification or with observed power. The record does not report its original statistical plan or sample-size rationale. Clear reporting of direct contrasts, analysis populations, missing-data handling and the endpoint’s direction would make the statistical findings easier to evaluate.

13. Limitations

• The posted summary statistics are the only data these results rest on; patient-level data, full protocols and statistical analysis plans are not available.
• Recalculated results are simple, unadjusted calculations from posted summaries. They do not replace the sponsor's prespecified analysis.

The main limitation of the posted statistical results is that the primary endpoint is described as a mean difference between tests, but the subtraction order is not stated. As a result, the negative estimates cannot be assigned an unambiguous direction of change. The registry record also does not define isotime beyond naming it as the assessment point, or explain how that point was selected.

Although the study has four arms, the posted results give arm-specific means and confidence intervals rather than direct between-arm contrasts. The record explicitly notes that two-arm comparisons for this continuous endpoint are missing or not estimable. No p-values or adjusted comparisons are listed. Consequently, the relative effects of high-flow versus low-flow delivery, or oxygen versus air, cannot be established from these results.

The primary outcome is described for 44 participants who completed the study, while enrollment is listed as 74. The registry record does not report participant flow, reasons for missing data, arm-specific completion information, or the method used to address missing observations. It also does not identify the analysis population in greater detail. The effect of these unknowns on the estimates cannot be determined.

Additional analytical details are absent, including the statistical model, treatment of repeated measurements, covariate adjustment, and any multiplicity strategy. The record does not report a clinically important difference threshold for the Modified Borg Scale. Adverse-event fields show zero values for both participants at risk and affected, so they do not provide an interpretable safety summary. These limitations constrain what can be concluded from the registry results.

14. Related Tutorials

15. Related Statistical Calculators

16. Sources

• ClinicalTrials.gov: High-Flow Oxygen in Reducing Shortness of Breath Caused by Exercise in Patients With Cancer — NCT02357134.
• Registry record last updated 2022-12-07; status last verified 2022-11.
• Results first posted 2021-06-01.
• Data retrieved from ClinicalTrials.gov on 2026-11-09.
• Group sizes, means, event counts and sponsor-reported p-values are reproduced as posted. Effect estimates, confidence intervals and p-values not attributed to the sponsor are independently calculated from the posted summary statistics.

17. Record Summary

NCT02357134 is a randomized, double-masked, parallel phase II trial sponsored by M.D. Anderson Cancer Center. It evaluates high-flow oxygen, high-flow air, low-flow oxygen and low-flow air during structured stationary bicycle exercise in patients with cancer and exertional dyspnea. Enrollment is listed as 74; the primary outcome description refers to 44 participants who completed the study. The study status is unknown.

The primary endpoint is the mean difference in Modified Borg Scale Dyspnea Intensity at isotime between the second exercise test, when all participants received low-flow oxygen, and the third test, when they received their assigned intervention. Arm-specific mean differences and 95% confidence intervals are posted, but direct two-arm continuous-outcome comparisons are missing or not estimable. The difference’s subtraction direction, analysis plan, missing-data handling and statistical rationale are not reported in the registry record.